Data

This section describes the preliminary data exploration prior to analysis, including assessment of distributions, evaluation of data quality and quantity, identification and removal of outliers, and selection of representative visualizations for the dataset.

Table 1. Summary of raw data table. Fungal species G. clavigera (Gc) and O. montium (Om) were exposed to five chemical treatments: no amendment, ethanol-only controls, and chemical profiles from British Columbia, Alberta, and Oregon. Fungal biomass (growth) and volatile emissions were measured. No-amendment and ethanol-only treatments were included as baseline controls for fungal growth and volatile emissions. 

First, I used histograms to assess whether fungal biomass was normally distributed within treatment groups. The data were not normally distributed within treatments, and several outliers were present, likely due to the small sample size (n = 10; Fig. 1). Because treatment groups are independent, non-parametric analyses were therefore used.

Distributions varied substantially among treatments, suggesting that monoterpene blends differentially affect fungal biomass. Biomass values for the British Columbia, control, and ethanol treatments were generally higher (clustered toward the right), whereas Alberta and Oregon treatments exhibited lower biomass (clustered toward the left) (Fig. 1). Given these distributional patterns, box plots are the most appropriate visualization for illustrating differences in fungal growth among treatments.

Figure 1. Distribution of fungal biomasses within each treatment group. Plots of different chemical amendments are segragated by black lines. Gc-Gc for G. clavigera culture (blue), Gc-Om for G. clavigera and O. montium co-culture (yellow), and Om-Om for O. montium only (red).

Figure 2. Diagnostic plots used to assess whether the distribution of fungal biomasses meet classical parametric assumptions. (a) The residuals versus fitted values plot indicates heteroscedasticity, with the variance in fungal biomass increasing across fitted values, suggesting unequal variances among treatment groups. (b) The Q–Q plot of standardized residuals shows clear deviations from the reference line in both tails, indicating non-normal residuals and the presence of outliers at both low and high biomass values.

Taken together, these diagnostics indicate violations of normality and homogeneity of variance assumptions. But since I am only interested in group differences and not effect sizes, the Kruskal-Wallis test was used to evaluate statistical difference. Dunn’s post-hoc test was subsequently applied for pairwise comparisons. 

Figure 3. Box plots displaying overall fungal biomass of G. clavigera and O. montium under different lodgepole pine defences. Lower case letters above box plots indicates significant groupings.

The Kruskal-Wallis test, followed by pairwise comparisons, revealed significant differences in fungal growth among treatments (Kruskal-Wallis test: χ² = 52.35, df = 2, p < 0.001). All amended treatments significantly suppressed fungal growth compared to both the control and ethanol. Although ethanol had a slight negative effect on growth, this difference was not statistically significant (Z = -1.18, p = 0.33). Since ethanol serves as the solvent for all amendments, it provides the appropriate baseline for assessing the effects of the treatments.

Next, I evaluated the normality of distribution of fungal volatile concentrations within each treatment group.  

Figure 4. Summary boxplots of fungal volatile concentrations. Fungal species G. clavigera (Gc) and O. montium (Om) were exposed to five chemical treatments: no amendment, ethanol-only controls, and chemical profiles from British Columbia, Alberta, and Oregon. The box plots show the distributions of volatile concentrations for each fungal × amendment combination. Dots at the ends of the whiskers represent outliers. 

Figure 4 shows that most volatile compounds exhibit little to no distinct response to treatment type. In contrast, isobutanol, cis-grandisol, and verbenone display elevated concentrations under specific amendments. These volatiles were therefore selected for further analysis. 

Figure 5. Summary residual vs fitted plots and Q-Q plots for distributions of each fungal volatile concentration. Plots are arranged side by side for each compound (total 8).

Figure 5 indicates that the distributions of most volatile compounds deviate from normality, largely due to the presence of outliers. cis-grandisol and verbenone exhibit particularly extreme outliers. Like the biomass data, volatile concentrations show heterogeneous variances across treatments and frequent outliers, indicating violations of parametric assumptions. Therefore, non-parametric analyses were applied, using Kruskal–Wallis tests followed by post-hoc pairwise comparisons to evaluate differences among treatment groups.